Don't anybody get too excited. But this shows how understanding of
biology is working its way into electronics, and at some point this
will become useful to us.
This is another article from one of my favorite places, Electronic
Engineering Times, copyright CMP Publications.
Vision chip's circuitry modeled on eye and brai
By R. Colin Johnso
BOUYGES, France -- Mimicking the human eye's neural networks and th
brain, an electronic "eye" on a chip can now handle tasks that rang
from reading sign language to avoiding collisions
The product of 10 years of research, the Generic Visual Perceptio
Processor (GVPP) was developed here by the Bureau Etudes Vision (BEV
Stockplus, said Patrick Pirim, the chip's inventor and the bureau'
principal scientist
"Now," said BEV Stockplus president Igor Marie De l'Isle, "we want t
bring this extraordinary chip to the U.S. market, because the detai
with which visual streams can be recognized enables new application
not possible before.
For instance, he said, sign language involves very detailed han
gestures that have so far been impervious to traditional recognitio
techniques. The company also claims to benefit traditiona
pattern-recognition applications, such as military target acquisitio
and fire control, as well as automobile-collision avoidance, adaptiv
cruise control and automatic "asleep-at-the-wheel" alarms
Input to the electronic eye can come from video, infrared or rada
signals. Real-time outputs perceive, recognize and analyze both stati
images and time-varying patterns for specific objects, their heading
speed, shading and color differences
By mimicking the eye plus the visual regions of the brain, the GVP
culls the essential features. So instead of capturing frames o
pixels, the chip identifies objects of interest, determines eac
object's speed and direction, then follows them by tracking thei
color through the scene
The chip emulates the eye, which has 5 million cones sensitive t
color, only 15 percent of which see "blue" (the rest are red an
green) and 140 million monochromatic rods that are 35 times mor
sensitive than the cones
The chip mimics the human eye's two processing steps, tonic an
phasic. Tonic processing auto-scales according to ambient ligh
conditions, enabling it to adapt to a range of luminosity. Phasi
processing determines movement by using local variables in feedbac
loops. As light's edges pass over the cones and rods of the eye, thes
local feedback loops detect contrast changes caused by objects movin
through the scene
For detecting smooth contours, rather than sharp contrast changes, th
eye adds ocular movement. The eye typically sweeps a scene about tw
to three times a second as well as making vibratory movements at abou
100 Hz. The faster jitter accounts for the visual acuity of th
eye--sensitivity to the smallest detectable feature, which is an edg
moving between two adjacent rods or cones
After all this processing, the visual signal is then sent to the brai
for higher-level observation and recognition tasks. Because onl
detected movement and color along with the shape and contour o
objects is sent up to the brain, rather than raw pixels, the averag
compression ratio of information, according to the company, is abou
145. For the future, Pirim is working on a "visual" mouse for
hand-gesture interface to computers that takes advantage of that hig
compression ratio
The electronic eye mimics the theoretical processing steps of the rea
eye with hard-wired silicon circuitry around each pixel in its senso
array. Each pixel is analyzed by the vision chip with hardware tha
determines and scales luminescence, tracks color, remembers movemen
in the previous moment, remembers the direction of previous movement
and deduces the speed of the detected objects from parallel phasic an
tonic neural circuitry
Basically, each parameter has an associated neuron that handles it
processing tasks in parallel. In addition, each pixel has tw
auxiliary neurons that define the zone in which an object i
located--that is, from the direction in which an object is moving
these neurons deduce the leading and trailing edge of the object an
mark with registers associated with the first (leading-edge) and las
(trailing-edge) pixel belonging to the object
Each of these silicon neurons is built with RAM, a few registers, a
adder and a comparator. The auxiliary neurons also have a multiplie
at their disposal
Supplied as a 100-pin module, the chip accommodates analog-input lin
levels for video input, with an input amplifier with programmable gai
autoscaling the signal
The modules measure 40 mm2, have 100 pins and can handle 20-MHz vide
signals. The chip is priced at $960. On a card with a socketed GVP
and 64 kbytes of Flash RAM the price comes to $1,500. A developer'
daughtercard version, with a motherboard accommodating inputs fo
video, a power supply and computer interface along with a PAL encode
and a library of software to manage the unit is available for $4,650
-- Lloyd Rasmussen
Senior Staff Engineer, Engineering Section
National Library Service for the Blind and Physically Handicapped
Library of Congress 202-707-0535
(work) lras@loc.gov www.loc.gov/nls/
(home) lras@sprynet.com
This archive was generated by hypermail 2b29 : Sun Dec 02 2012 - 01:30:04 PST